annotate rename.py @ 0:65c9ce351343 draft

planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
author yating-l
date Fri, 20 Jan 2017 17:06:52 -0500
parents
children 0d13e4410c3d
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
1 """
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
2 Call rename to rename scaffolds in reference genome so that the sequence names are less than 31 characters. Rename all scaffolds to scaffold_1, scaffold_2, ..., scaffold_N and provide a name mapping file
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
3 """
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
4 import sys
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
5 from collections import OrderedDict
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
6
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
7 def rename(inputfile, outputfile, indexfile):
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
8 namemap = OrderedDict()
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
9 with open(outputfile, 'w') as out:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
10 with open(inputfile, 'r') as rf:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
11 lines = rf.readlines()
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
12 i = 1
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
13 for line in lines:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
14 if ">" in line:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
15 oldname = line[1:].rstrip()
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
16 newname = "scaffold_" + str(i)
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
17 line = ">" + newname
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
18 i = i+1
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
19 namemap[oldname] = newname
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
20 #TODO: Add line breaks to chromosome sequences that are in a single line
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
21 #else:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
22 #if (len(line) > 50):
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
23 #for
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
24 out.write(line.rstrip() + "\n")
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
25 with open(indexfile, 'w') as index:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
26 for k in namemap:
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
27 index.write(k + "\t" + namemap[k] + "\n")
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
28
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
29 def main():
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
30 inputfile = str(sys.argv[1])
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
31 outputfile = str(sys.argv[2])
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
32 indexfile = str(sys.argv[3])
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
33 rename(inputfile, outputfile, indexfile)
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
34
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
35 if __name__ == "__main__":
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
36 main()
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
37
65c9ce351343 planemo upload commit af6c4dc7f6d6795f2b2db7299eb14ae1d5622ac1
yating-l
parents:
diff changeset
38